To evaluate the psychometric properties (Inter-rater reliability, measurement error, construct validity, criterion predictive validity, responsiveness, floor and ceiling effects) of the mILOA in acutely hospitalised, older general medical patients.
Design
A prospective observational study with psychometric analysis was conducted in general medical patients aged over 65 years who were referred to physiotherapy at two Australian hospitals.
Methodological Approach
Inter-rater reliability (intraclass correlation coefficient (ICC)) and measurement error (Standard Error of Measurement (SEM) and Minimal Detectable Change, MDC95) were calculated using two independently completed mILOA assessments within 24 hours. Construct validity was established by correlating the mILOA with the de Morton Mobility Index (DEMMI), a valid and reliable mobility measure in older hospitalised people (convergent validity, Spearman’s rho), and comparing mILOA scores in patients who were discharged home or “not home” (known-groups validity, Mann-Whitney U test). Criterion predictive validity (Area Under the Receiver Operating Curve, AUC) was determined using the admission mILOA and discharge destination. Responsiveness (effect size from admission to discharge), floor and ceiling effects, and mILOA completion time were also examined.
A total of 246 participants were included (median age 83.2 [Interquartile Range 46.10–88.00] years, 47% female).The median mILOA on admission was 12.00 [7.00-19.00] and 8.00 [5.00-13.00] at discharge; length of hospital stay was 7.00 [3.00-12.00] days.
The inter-rater reliability of the mILOA was excellent (ICC 0.88 (95% CI 0.79-0.93), with a variability due to measurement error of 2.36 and MDC95 of 6.53 points. The mILOA scores correlated significantly and positively with DEMMI scores (Spearman’s rho= 0.86 (95% CI 0.81-0.90), p0.001) and it could differentiate between participants who were discharged home (median 7.00 [4.00-12.00]) and those who were not discharged home (median 15.00 [9.00-22.00], p0.001). The mILOA may also be able to predict discharge destination (AUC 0.74 (95% CI 0.66-0.81)). No floor or ceiling effects were observed and the mILOA was highly responsive to change (effect size 0.80). It took a median of 12.5 [8.0-18.0] minutes to complete the mILOA.
The mILOA has excellent inter-rater reliability, excellent construct validity (convergent validity and known group validity), may be able to predict discharge home (criterion predictive validity), is responsive to change and has no floor or ceiling effects.
The mILOA is a relatively quick, valid, reliable and responsive outcome measure to assess mobility in acutely hospitalised older general medical inpatients. Further research may be needed to examine the structural validity of the mILOA using item response theory approaches such as Rasch analysis to ensure that the items measure the same underlying construct.
Psychometric studies
Mobility
